diff --git a/src/lerobot/datasets/pipeline_features.py b/src/lerobot/datasets/pipeline_features.py
index cf02a52ac..659c04405 100644
--- a/src/lerobot/datasets/pipeline_features.py
+++ b/src/lerobot/datasets/pipeline_features.py
@@ -70,19 +70,21 @@ def aggregate_pipeline_dataset_features(
     initial_features: dict[PipelineFeatureType, dict[str, Any]],
     *,
     use_videos: bool = True,
+    exclude_images: bool = False,
     patterns: Sequence[str] | None = None,
 ) -> dict[str, dict]:
     """
     Aggregates and filters pipeline features to create a dataset-ready features dictionary.
 
     This function transforms initial features using the pipeline, categorizes them as action or observations
-    (image or state), filters them based on `use_videos` and `patterns`, and finally
+    (image or state), filters them based on `exclude_images` and `patterns`, and finally
     formats them for use with a Hugging Face LeRobot Dataset.
 
     Args:
         pipeline: The DataProcessorPipeline to apply.
         initial_features: A dictionary of raw feature specs for actions and observations.
-        use_videos: If False, image features are excluded.
+        use_videos: Storage dtype for image features that are kept. If True, images are stored as "video", if False, they are stored as "image".
+        exclude_images: If True, image features are dropped entirely from the output.
         patterns: A sequence of regex patterns to filter action and state features.
                   Image features are not affected by this filter.
 
@@ -120,7 +122,7 @@ def aggregate_pipeline_dataset_features(
             )
 
             # 2. Apply filtering rules.
-            if is_image and not use_videos:
+            if is_image and exclude_images:
                 continue
             if not is_image and not should_keep(key, compiled_patterns):
                 continue
diff --git a/tests/processor/test_pipeline.py b/tests/processor/test_pipeline.py
index 57e948279..0e9746a63 100644
--- a/tests/processor/test_pipeline.py
+++ b/tests/processor/test_pipeline.py
@@ -2370,14 +2370,32 @@ def test_aggregate_images_when_use_videos_false():
     out = aggregate_pipeline_dataset_features(
         pipeline=rp,
         initial_features={PipelineFeatureType.ACTION: {}, PipelineFeatureType.OBSERVATION: initial},
-        use_videos=False,  # expect "image" dtype
+        use_videos=False,  # images kept, stored as "image" dtype
         patterns=None,
     )
 
     key = f"{OBS_IMAGES}.back"
     key_front = f"{OBS_IMAGES}.front"
-    assert key not in out
-    assert key_front not in out
+    assert key in out
+    assert key_front in out
+    assert out[key]["dtype"] == "image"
+    assert out[key_front]["dtype"] == "image"
+    assert out[key]["shape"] == initial["back"]
+
+
+def test_aggregate_images_excluded():
+    rp = DataProcessorPipeline([AddObservationStateFeatures(add_front_image=True)])
+    initial = {"back": (480, 640, 3)}
+
+    out = aggregate_pipeline_dataset_features(
+        pipeline=rp,
+        initial_features={PipelineFeatureType.ACTION: {}, PipelineFeatureType.OBSERVATION: initial},
+        exclude_images=True,
+        patterns=None,
+    )
+
+    assert f"{OBS_IMAGES}.back" not in out
+    assert f"{OBS_IMAGES}.front" not in out
 
 
 def test_aggregate_images_when_use_videos_true():